Non-Linear Stationary Subspace Analysis with Application to Video Classification

Mahsa Baktashmotlagh, Mehrtash Harandi, Abbas Bigdeli, Brian Lovell, Mathieu Salzmann
; Proceedings of the 30th International Conference on Machine Learning, PMLR 28(3):450-458, 2013.

Abstract

Low-dimensional representations are key to the success of many video classification algorithms. However, the commonly-used dimensionality reduction techniques fail to account for the fact that only part of the signal is shared across all the videos in one class. As a consequence, the resulting representations contain instance-specific information, which introduces noise in the classification process. In this paper, we introduce Non-Linear Stationary Subspace Analysis: A method that overcomes this issue by explicitly separating the stationary parts of the video signal (i.e., the parts shared across all videos in one class), from its non-stationary parts (i.e., specific to individual videos). We demonstrate the effectiveness of our approach on action recognition, dynamic texture classification and scene recognition.

Cite this Paper


BibTeX
@InProceedings{pmlr-v28-baktashmotlagh13, title = {Non-Linear Stationary Subspace Analysis with Application to Video Classification}, author = {Mahsa Baktashmotlagh and Mehrtash Harandi and Abbas Bigdeli and Brian Lovell and Mathieu Salzmann}, booktitle = {Proceedings of the 30th International Conference on Machine Learning}, pages = {450--458}, year = {2013}, editor = {Sanjoy Dasgupta and David McAllester}, volume = {28}, number = {3}, series = {Proceedings of Machine Learning Research}, address = {Atlanta, Georgia, USA}, month = {17--19 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v28/baktashmotlagh13.pdf}, url = {http://proceedings.mlr.press/v28/baktashmotlagh13.html}, abstract = {Low-dimensional representations are key to the success of many video classification algorithms. However, the commonly-used dimensionality reduction techniques fail to account for the fact that only part of the signal is shared across all the videos in one class. As a consequence, the resulting representations contain instance-specific information, which introduces noise in the classification process. In this paper, we introduce Non-Linear Stationary Subspace Analysis: A method that overcomes this issue by explicitly separating the stationary parts of the video signal (i.e., the parts shared across all videos in one class), from its non-stationary parts (i.e., specific to individual videos). We demonstrate the effectiveness of our approach on action recognition, dynamic texture classification and scene recognition.} }
Endnote
%0 Conference Paper %T Non-Linear Stationary Subspace Analysis with Application to Video Classification %A Mahsa Baktashmotlagh %A Mehrtash Harandi %A Abbas Bigdeli %A Brian Lovell %A Mathieu Salzmann %B Proceedings of the 30th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2013 %E Sanjoy Dasgupta %E David McAllester %F pmlr-v28-baktashmotlagh13 %I PMLR %J Proceedings of Machine Learning Research %P 450--458 %U http://proceedings.mlr.press %V 28 %N 3 %W PMLR %X Low-dimensional representations are key to the success of many video classification algorithms. However, the commonly-used dimensionality reduction techniques fail to account for the fact that only part of the signal is shared across all the videos in one class. As a consequence, the resulting representations contain instance-specific information, which introduces noise in the classification process. In this paper, we introduce Non-Linear Stationary Subspace Analysis: A method that overcomes this issue by explicitly separating the stationary parts of the video signal (i.e., the parts shared across all videos in one class), from its non-stationary parts (i.e., specific to individual videos). We demonstrate the effectiveness of our approach on action recognition, dynamic texture classification and scene recognition.
RIS
TY - CPAPER TI - Non-Linear Stationary Subspace Analysis with Application to Video Classification AU - Mahsa Baktashmotlagh AU - Mehrtash Harandi AU - Abbas Bigdeli AU - Brian Lovell AU - Mathieu Salzmann BT - Proceedings of the 30th International Conference on Machine Learning PY - 2013/02/13 DA - 2013/02/13 ED - Sanjoy Dasgupta ED - David McAllester ID - pmlr-v28-baktashmotlagh13 PB - PMLR SP - 450 DP - PMLR EP - 458 L1 - http://proceedings.mlr.press/v28/baktashmotlagh13.pdf UR - http://proceedings.mlr.press/v28/baktashmotlagh13.html AB - Low-dimensional representations are key to the success of many video classification algorithms. However, the commonly-used dimensionality reduction techniques fail to account for the fact that only part of the signal is shared across all the videos in one class. As a consequence, the resulting representations contain instance-specific information, which introduces noise in the classification process. In this paper, we introduce Non-Linear Stationary Subspace Analysis: A method that overcomes this issue by explicitly separating the stationary parts of the video signal (i.e., the parts shared across all videos in one class), from its non-stationary parts (i.e., specific to individual videos). We demonstrate the effectiveness of our approach on action recognition, dynamic texture classification and scene recognition. ER -
APA
Baktashmotlagh, M., Harandi, M., Bigdeli, A., Lovell, B. & Salzmann, M.. (2013). Non-Linear Stationary Subspace Analysis with Application to Video Classification. Proceedings of the 30th International Conference on Machine Learning, in PMLR 28(3):450-458

Related Material